os-memory-manager

Automate agent memory management by applying a phase-driven lifecycle that captures session data, promotes key findings to long-term memory, and archives logs. This enforces consistent memory hygiene and auditable persistence without manual curation.

Promote session findings to long-term memory using safety gates that prompt user confirmation. This prevents duplication and conflicting facts, moving reviewed data from per-session logs into a curated long-term memory file.

A three-tier memory model structures agent data into Auto-memory, Long-term memory, and per-session logs. It separates immediate context from permanent knowledge and daily records, enabling consistent data governance and auditability.

Yes, you need Python-based tooling and a running OS init state to perform memory capture, entitlement gating, and archival actions. These prerequisites enable the automated lifecycle management of agent memory.

Memory hygiene requires safety gates to prompt user confirmation before promoting findings. This precaution prevents duplication and conflicting facts, ensuring data governance and auditable memory persistence across sessions.

Yes, you can archive per-session logs without data loss. The system separates daily session logs from promoted long-term memory, ensuring key findings are preserved while daily records are archived for auditability.